Content uploaded by Michael A Perelman
Author content
All content in this area was uploaded by Michael A Perelman on Feb 04, 2022
Content may be subject to copyright.
Sexual Medicine
Development and Validation of a Premature Ejaculation
Diagnostic Tool
Tara Symonds
a,
*, Michael A. Perelman
b
, Stanley Althof
c
, Franc¸ois Giuliano
d
,
Mona Martin
e
, Kathryn May
a
, Lucy Abraham
a
, Anna Crossland
a
, Mark Morris
a
a
Pfizer Ltd, Sandwich, Kent, UK
b
NY Presbyterian, Weill Medical College of Cornell University, New York, NY, USA
c
Case Western Reserve University School of Medicine, Cleveland, OH, USA
d
AP-HP, Raymond Poincare Hospital Garches, Garches, France
e
HRA Inc, Seattle, WA, USA
european urology 52 (2007) 565–573
available at www.sciencedirect.com
journal homepage: www.europeanurology.com
Article info
Article history:
Accepted January 5, 2007
Published online ahead of
print on January 16, 2007
Keywords:
Diagnosis
Premature ejaculation
Diagnostic tool
Orgasmic dysfunction
Rapid ejaculation
Sexual dysfunction
Questionnaire
Score
Abstract
Objectives: Diagnosis of premature ejaculation (PE) for clinical trial purposes has
typically relied on intravaginal ejaculation latency time (IELT) for entry, but this
parameter does not capture the multidimensional nature of PE. Therefore, the
aim was to develop a brief, multidimensional, psychometrically validated instru-
ment for diagnosing PE status.
Methods: The questionnaire development involved three stages: (1) Five focus
groups and six individual interviews were conducted to develop the content;
(2) psychometric validation using three different groups of men; and (3) genera-
tion of a scoring system. For psychometric validation/scoring system develop-
ment, data was collected from (1) men with PE based on clinician diagnosis, using
DSM-IV-TR, who also had IELTs 2 min (n= 292); (2) men self-reporting PE
(n= 309); and (3) men self-reporting no-PE (n= 701).
Standard psychometric analyses were conducted to produce the final ques-
tionnaire. Sensitivity/specificity analysis was used to determine an appropriate
scoring system.
Results: The qualitative research identified 9 items to capture the essence of
DSM-IV-TR PE classification. The psychometric validation resulted in a 5-item,
unidimensional, measure, which captures the essence of DSM-IV-TR: control,
frequency, minimal stimulation, distress, and interpersonal difficulty. Sensitiv-
ity/specificity analyses suggested a score of 8 indicated no-PE, 9 and 10 probable
PE, and 11 PE.
Conclusions: The development and validation of this new PE diagnostic tool has
resulted in a new, user-friendly, and brief self-report questionnaire for use in
clinical trials to diagnose PE.
#2007 European Association of Urology. Published by Elsevier B.V. All rights reserved.
* Corresponding author. Worldwide Outcomes Research (IPC160),
Pfizer Global Research and Development, Ramsgate Road, Sandwich, Kent,
CT13 9NJ, United Kingdom. Tel. +44 1304 641553; Fax: +44 1304 658823.
E-mail address: tara.symonds@pfizer.com (T. Symonds).
0302-2838/$ – see back matter #2007 European Association of Urology. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.eururo.2007.01.028
1. Introduction
Premature ejaculation (PE) is the most common
ejaculatory disorder. Epidemiologic studies indicate
that PE has an estimated prevalence of approxi-
mately 16–38% across all age groups of the male
population [1].
Although defined in several ways, the most
widely accepted definition is Diagnostic and Statistical
Manual of Mental Disorders, Fourth Edition, Text Revision
(DSM-IV-TR) [2], which uses vague terms such as
‘‘minimal sexual stimulation,’’ ‘‘before the person
wishes it,’’ and ‘‘causes marked distress or inter-
personal difficulty.’’
Traditionally, intravaginal ejaculation latency
time (IELT) has served in clinical trials as an opera-
tional marker for DSM-IV-TR criteria. Frequently,
the IELT inclusion criterion has been set at 1 or
2 min. Waldinger et al [3] found, in their meta-
analysis of PE drug treatment studies, of 41 studies
that used a time-based operational definition 30
used 1 or 2 min as the PE definition. Three minutes
or less was used in 8 studies, and 30 s, 4 min, and
5 min were used in 1 study each. However, the 1- or
2-min cut point does not fit all. There are men who
would still consider themselves premature with an
IELT of 3, 4, or even 5 min, as evidenced by Patrick
et al’s [4] study. Conversely, there are men who do
not consider they have PE with IELTs of 1 or 2 min.
The International Consultation on Sexual Dysfunc-
tions ejaculation committee members [5,6] sug-
gested that IELT, control, and distress elements all
be considered when diagnosing PE, because PE is not
simply a function of time; men also complain of lack
of control, sexual dissatisfaction, and/or concern
about partner dissatisfaction.
Since diagnosis seems to warrant more than just
an assessment of time, it was proposed that a self-
report measure that encompasses the essence of
DSM-IV-TR be developed.
2. Part 1: Pilot tool development
2.1. Method
2.1.1. Study design
Five focus groups (6–7 men each) and six individual
interviews were convened with men who self-
reported having PE, or who were currently receiving
treatment for their PE. The focus groups were
conducted in the United States and Germany.
Because we were advised against use of a focus
group model in Spain, we conducted individual
interviews. Focus groups and interviews were
conducted in subjects’ native language. These
countries were chosen to give a range of cultural
experience. Following this qualitative data collec-
tion, a panel of five clinical experts was then
convened in July 2004 to add input to the process.
The panel members were chosen on the basis of
geographic distribution (France, Australia, United
States), years of experience in diagnosing PE, and to
multidisciplinary variability (sex therapist, primary
care provider, urologist, experimental psychologist).
2.1.2. Study population
2.1.2.1. Focus groups and individual interviews. Recruit-
ment was based on a diagnosis by a physician or
by self-report. Participation in interviews conducted
in Spain was based solely on a man’s self-report of
experiencing PE.
The average age of the men was 44.34 yr (standard
deviation = 11.44). Regardless of marital status, a
current relationship was required for participation;
most (62%) were married. Average years spent in
their current relationship was just under 10. Nearly
40% had middle-class incomes, 65% had at least
some college education, and the group was 72%
Caucasian. The average number of years with PE
was just under 8, and the men had either lifelong or
acquired PE.
2.1.3. Study procedure
2.1.3.1. Focus groups/individual interviews/clinical expert
panel. After informed consent was obtained, partic-
ipants were asked 16 open-ended questions about
their experience with PE and its impact on their
relationship. Example questions included: How
would you describe PE? How do you think your
experience of ejaculation compares with what you
might expect to be a ‘‘normal’’ or usual sexual
experience? How does having PE make you feel
(distressed, angry, frustrated, disappointed, anno-
yed, bothered)? Is ‘‘control’’ an issue for you?
For the expert panel, clinical experts were asked
to present and discuss their clinical experience with
PE, focusing in particular on the process of diag-
nosis.
2.1.4. Results
The results from the qualitative work and expert
panel are summarised below:
When men described their symptoms, they talked
about problems with timing of ejaculation, rather
than lasting a certain amount of time.
They generally rejected the concept that PE could
be defined by absolute time to ejaculation,
particularly as their usual sexual practice tended
european urology 52 (2007) 565–573566
to incorporate other sexual activities. It was
unusual for many of them to have a period of
uninterrupted intercourse culminating in ejacula-
tion.
They identified control over ejaculation as being
important, but this tended to be discussed as a
better degree of continence, rather than absolute
control, which they did not see as realistic.
Most men did not identify with the word ‘‘dis-
tress’’ to describe the emotional impact of their PE.
Rather, the word ‘‘frustration’’ seemed to be the
most appropriate and widely used descriptor.
Components men wanted to see incorporated
into a diagnostic measure were concepts such as:
Ejaculating before you wish to
Inability to delay ejaculation
Inability to control ejaculation
The 9-item PEDT was tested for understanding in
two additional focus groups (United States and
Germany). There were no issues with the content; a
definition of PE was added for clarity. The version in
Appendix A was taken forward for validation.
3. Part 2: validation
3.1. Method
3.1.1. Study design
We recruited a cohort of men with PE defined
according to DSM-IV-TR and known IELT of 2 min
in 70% of coital attempts. These men (all from the
US) were asked to use a stopwatch to time IELT over
a 4-wk duration (time-defined population).
The self-reported population (self-reported PE
and self-reported no-PE) was recruited by using
Harris Interactive Service Bureau (HISB), a US Web-
based survey system. The incentive for being on this
panel was a system of points awarded for surveys
completed, which could then be redeemed at an
online store.
3.1.2. Study populations
3.1.2.1. Time-defined population. In total, 292 men were
recruited, with known IELTs of than 2 min in >70%
of coital attempts.
3.1.2.2. Self-reported population. We categorized as ‘‘self-
reported PE’’ (n= 309) subjects who responded with
‘‘probably yes’’ or ‘‘definitely yes’’ to the following
question: ‘‘Premature ejaculation is a term com-
monly used to refer to the condition when a man
ejaculates before he and/or his partner wants him
to. Given this definition, do you think you suffer
from premature ejaculation?’’ Also, these men had
to have partners who did not have any sexual
problems.
Subjects who responded with either ‘‘definitely
not’’ or ‘‘not sure’’ were categorized as ‘‘self-reported
no-PE’’ (n=701).
3.1.3. Demographics
The demographics for each of the cohorts can be
seen in Table 1.
3.1.4. Study procedure
3.1.4.1. Time-defined population. Men were recruited by
advertisement and were screened with the use
of the following criteria: experienced PE for at least
10 yr, aged 18–65 yr, in a stable relationship for at
least the last 6 mo, fulfilled DSM-IV-TR criteria for
PE, and reported no erectile dysfunction (according
to the International Index of Erectile Function [7]).
Men meeting the initial screening criteria completed
a consent form, provided initial questionnaire
Table 1 – Demographics for the time-defined, self-reported PE, and self-reported no-PE cohorts
Population Time-defined (n= 292) Self-reported PE (n= 309) Self-reported no-PE (n= 701)
Age (yr)
18–41 72% 33% 43%
42–65 28% 67% 57%
Relationship
Yes 100%
a
74% 70%
>1 year 76% 94% 91%
IELT
Mean (SE) 66 s (1.78) 279.4 s (19.22) 490.9 s (36.26)
Median 65 s 180 300
PE = premature ejaculation; IELT = intravaginal ejaculation latency time; SE = standard error.
a
All men in the time-defined population had to be in a stable relationship for at least 6 mo.
european urology 52 (2007) 565–573 567
information (demographics and PEDT), and were
given an explanation of the 4-wk diary process. The
men were then given an electronic diary with
embedded stopwatch to take home to record IELTs
and confirm an IELT of 2 min in 70% of coital
attempts. Each subject was given the option of either
he or his partner operating the stopwatch. Which-
ever partner timed the ejaculation on the first
occasion continued to do so throughout the course
of the 4-wk study. Two weeks after subjects were
enrolled, they received a second PEDT by mail, along
with a self-addressed stamped envelope, to com-
plete and return (for assessing retest reliability). At
the end of the 4-wk period, participants returned
and completed a further PEDT survey.
3.1.4.2. Self-reported population. A survey asking screen-
ing criteria similar to those above was sent out to a
general population of men; it also contained the
PEDT and a question about PE status (as given above)
to confirm the subjects’ belief that they ejaculated
prematurely or did not ejaculate prematurely. IELT
information was not captured for this group because
it was meant to be broad and not constrained by an
IELT cut point.
3.1.5. Statistical analyses
3.1.5.1. Psychometric analyses. Various analyses were
carried out to identify items to include in the final
PEDT (Table 2).
Once the final items for the tool were selected, the
resulting tool was then checked to see that internal
consistency, test-retest reliability, and known-
groups validity still applied.
The psychometric tests listed in Table 2 were
also performed on the self-reported PE population
to explore whether a similar set of items was
selected.
Before agreeing on the final content of the tool,
the psychometric test results were reviewed by
three experts (Drs S. Althof [S.A.], F. Giuliano [F.G.],
M. Perelman [M.P.]).
3.2. Results
3.2.1. Performance of individual items
Item-to-item correlations were acceptable. The
correlation coefficient between items 9 and 4, 5, 7,
and 8 was greater than the predefined critical value
of 0.70 (see Appendix A for item content). The overall
Cronbach alpha for the 9-items was 0.86. Test-retest
reliability was good (0.82), and all items statistically
significantly discriminated between those with PE
and those without PE.
One factor was retained (Eigen value of the first
factor: 4.03), which explained 95% of the variance
and factor loadings >0.4; loadings ranged between
0.41 (item 6) and 0.88 (item 9).
3.2.2. Reduction of the items
Following a discussion with the three experts, items
1, 2, 3, 4, and 8 were retained (see Appendix A).
Although the known-groups validity for each item
was very good, item 6 performed less well at
discriminating between the two populations than
the other eight items. Item 9 failed the prespecified
criterion of an item-to-item correlation of <0.70 with
item 7. It was also highly correlated with items 4, 5,
and 8. Item 7 was highly correlated with items 4 and
5. In addition, there was a high correlation between
items 4 and 5. Items 7 and 9 were deemed redundant
and were removed. Similarly, there was some
redundancy between items 4 and 5; item 5 was
removed.
In terms of clinical face validity, the five selected
items best summarized the essence of DSM-IV-TR.
Items 1–3 covered the concept of control, but 2 and 3
also addressed frequency and minimal sexual
stimulation, respectively. Item 4 addressed the
concept of distress, and item 8 covered the issue
of interpersonal difficulty.
3.2.3. Psychometrics of the 5-item PEDT
Factor analysis of the 5 items retained the one-factor
solution.
The reliability of the 5-item version was good
(Cronbach alpha = 0.71; test-retest reliability = 0.73).
Difference in mean score between the time-defined
population and self-reported no-PE group was
Table 2 – Psychometric analyses
Factor analysis Common factor analysis with Promax
rotation [8]. A priori criteria for
domain identification and item
retention were (1) Eigen values >1.0;
(2) items with factor loadings >0.4
Item-to-item correlation Pearson correlation coefficient was
used; correlation coefficients >0.70
indicate redundancy.
Reliability
Internal consistency Cronbach alpha score 0.7 is
considered good internal consistency.
Test-retest reliability Intraclass correlation coefficient (ICC):
the minimal acceptable level was
defined as 0.70.
Validity
Known-groups validity Independent ttest to determine
whether items could discriminate
between those with PE and those
without PE.
PE = premature ejaculation.
european urology 52 (2007) 565–573568
highly statistically significantly different (mean
difference = 11.8; 95% confidence interval, 11.3–
12.2); p<0.0001).
The psychometrics from the self-reported PE
population confirmed that items 1, 2, 3, 4, and 8
were the best items to retain.
The final version of the tool (Appendix B) was
taken forward for assessment of the most appro-
priate cut score to indicate presence or absence of
PE.
4. Part 3: scoring system
4.1. Method
4.1.1. Sample size
The number of men in the time-defined population
was determined by a general formula for sample
size estimation when constructing a two-sided
confidence interval for single test accuracy on
sensitivity [9].
The minimum number of subjects required for
each cohort (PE and no-PE) was 246.
4.1.2. Sensitivity/specificity analysis
Sensitivity was defined as the proportion of subjects
with the disorder who were diagnosed as having the
disorder (true positive rate); specificity was defined
as the proportion of subjects without the disorder
who were diagnosed as not having the disorder (true
negative rate).
The time-defined population was the source of
information for subjects ‘‘with’’ the disorder. The
no-PE population was used as the group ‘‘without’’
the disorder. The scoring system for the diagnostic
tool was derived from the point at which the
sensitivity/specificity ratio was closest to unity
(this approach maximized both sensitivity and
specificity).
The primary analysis was conducted on the time-
defined PE population and the self-reported no-PE
population. Secondary analysis used the self-
reported PE population and the self-reported no-PE
population.
4.2. Results
4.2.1. Primary analysis: time-defined PE versus self-reported
no-PE (definitely no/not sure)
The diagnostic tool discriminated between the two
cohorts (sample size: 294 vs. 701, respectively)
extremely well, suggesting that both groups were
answering the questions very differently. When
sensitivity/specificity were maximised, a score
between 10 and 11 was the best predictor of PE
status.
4.2.2. Secondary analysis: self-reported PE (definitely yes
and probably yes) versus self-reported no-PE (definitely no
and not sure)
There was a good fit between the two self-reported
cohorts (sample size: 309 vs. 701, respectively). A
score of 7 indicated no-PE and 8 indicated PE.
4.2.3. Additional exploratory analyses
The time-defined versus self-reported no-PE (‘‘defi-
nitely no’’ only, n= 474) had the strongest fit; the two
cohorts clearly responded very differently. The
sensitivity/specificity analysis produced a score of
8 for no-PE and 9 for PE.
The final analysis used men who self-reported PE
as ‘‘definitely yes’’ only (n= 68) versus the full data
set of men who self-reported no-PE (‘‘definitely no’’
and ‘‘not sure’’). A score of 9 indicated no PE and
10 indicated PE.
The data were reviewed and discussed with the
experts. The scoring system was set at 11 to
suggest a diagnosis of PE. Since the predicted cut
point between PE and no-PE status went as low as
8, and the ‘‘not sure’’ group would have a score
between 9 and 10, it was agreed that ‘‘probable PE’’
should be defined in this range; any man scoring a 9
or 10 would necessitate further assessment. A score
below 9 would indicate low likelihood of PE.
5. Discussion
A literature search revealed only one tool that
purported to discriminate PE from non-PE: Chinese
Index of Premature Ejaculation [10]. However, it was
developed as an efficacy measure and did not
specifically address the DSM-IV-TR criteria. This
study’s objective was to develop a validated, short
tool using DSM-IV-TR classification criteria to
diagnose PE. While individual clinicians rely on
the DSM-IV-TR criteria, there is wide variability in
application. The PEDT was developed to standardise
the diagnosis of PE in future clinical trials and was
designed to capture the main elements of DSM-IV-
TR: control, frequency, minimal sexual stimulation,
distress, and interpersonal difficulty. These ele-
ments mirrored the patient concerns arising in focus
groups and individual interviews. In fact, from the
focus groups, it was clear that PE’s impact on
men was primarily one of lack of control and
emotional impact on the man and his partner.
These concerns accounted for the large number of
emotional concepts incorporated into the draft tool
european urology 52 (2007) 565–573 569
to determine which emotional descriptor was
relevant to most men (frustration, lack of confi-
dence, dissatisfaction, and disappointment). Feel-
ings of lack of confidence were previously shown to
be the most common complaint associated with PE
[11]. However, in attempting to capture the essence
of distress it was the word ‘‘frustration’’ that most
resonated with the men interviewed. Some men felt
that ‘‘distress’’ was more a female-oriented concept,
and others believed this term was too strong for
what they felt about their condition. These same
views had been previously cited by men in other
focus groups during the development of an outcome
measure: Index of Premature Ejaculation [12];
‘‘distress’’ is now defined as ‘‘frustration’’ in this
measure. The psychometric analyses confirmed this
item best captured men’s distress/emotional impact
of experiencing PE. Also, Symonds et al [11] high-
lighted that men were aware of the impact of PE on
their relationship, showing that such an item in any
diagnostic tool is very relevant.
Overall, the final combination of items showed
excellent psychometric properties supporting its
validity and reliability. The subsequent sensitivity/
specificity analyses suggested a consistent cut score
of 8 ‘‘no PE,’’ 9 and 10 ‘‘probable PE,’’ and 11 ‘‘PE.’’
This result was obtained despite the fact that the
self-reported PE men were older than the time-
defined PE men and was collected via an Internet
panel. Using an Internet system to select the self-
reported PE population could have resulted in a quite
dissimilar group; however, educational background,
ethnicity, and relationship history were all similar.
The difference observed in IELTs was expected as
Waldinger et al’s [13] earlier research showed self-
reported estimates of IELT tend to be higher than
those recorded using a stopwatch. While age differ-
ence was not expected, the younger age may have
been driven by the fact that men had to take home an
electronic diary to record IELT, and that younger men
may be more confident in usingelectronics. However,
the wide spectrum of ages and the fact that the time-
defined/self-reported PE populations had very dif-
ferent IELTs strengthen the PEDT’s applicability
across a broad group of men with PE.
The PEDT will help overcome the variability of
application of DSM-IV-TR criteria and allow clinicians
to have a sense of assurance that a patient merits
diagnosis and treatment. The clinician, whose
patient scores ‘‘probable PE,’’ will be appropriately
prompted to investigate further to determine actual
PE status.
In the absence of a gold standard for diagnosing
PE and evidence of variation in the relationship
between absolute time and PE, the next step for
testing the PEDT’s validity will be to test it against
expert clinician diagnosis. This step will allow
consideration of any danger of ‘‘overdiagnosis’’ by
the tool. Since we know that IELT is culturally
dependent [14], the PEDT’s cross-cultural applica-
bility must also be researched further.
6. Conclusions
The PEDT is a short, psychometrically validated
measure that can be easily administered to facilitate
the diagnosis of PE.
Conflicts of interest
Tara Symonds, Kathryn May, Lucy Abraham, Anna
Crossland, and Mark Morris are all Pfizer Inc
employees.
Michael Perelman, Stanley Althof, and Franc¸ois
Giuliano are paid consultants to Pfizer Inc.
Mona Martin has no conflict of interest.
european urology 52 (2007) 565–573570
Appendix A. Draft Diagnostic Tool
european urology 52 (2007) 565–573 571
References
[1] Spector IP, Carey MP. Incidence and prevalence of sexual
dysfunctions: a critical review of the empirical literature.
Arch Sex Behav 1990;19:389–408.
[2] Diagnostic and statistical manual of mental disorders,
fourth edition, text revision: DSM-IV-TR. Washington
(DC): American Psychiatric Association; 2000. p. 554.
[3] Waldinger MD, Zwinderman AH, Schweitzer DH, et al.
Relevance of methodological design for the interpretation
of efficacy of drug treatment of premature ejaculation: a
systematic review and meta-analysis. Int J Imp Res
2004;1:1–13.
[4] Patrick DP, Althof SE, Pryor JL, et al. Premature ejaculation:
an observational study of men and their partners. J Sex
Med 2005;2:358–67.
Appendix B. Final PE Diagnostic Tool
european urology 52 (2007) 565–573572
[5] McMahon C, Abdo C, Incrocci L, et al. Disorders of orgasm
and ejaculation in men. J Sex Med 2004;1:58–65.
[6] Rowland DL. Treatment of premature ejaculation: select-
ing outcomes to determine efficacy. Bulletin Int Soc Sex
Impot Res 2003;10:26–7.
[7] Rosen RC, Riley A, Wagner G, et al. The International
Index of Erectile Function: a multidimensional scale for
the assessment of erectile dysfunction. J Urol 1997;49:
822–30.
[8] Cappelleri J, Gerber R. In: Chow SC, editor. Exploratory
factor analysis in Encyclopedia of Biopharmaceutical Sta-
tistics. New York: Marcel Dekker; 2003. p. 343–8.
[9] Zhou X-H, Obuchowski NA, McClish DK. Statistical meth-
ods in diagnostic medicine. New York: John Wiley and
Sons; 2002. p. 196–8.
[10] Yuan JM, Xin ZC, Jiang H, et al. Sexual function of prema-
ture ejaculation patients assessed with the Chinese Index
of Premature Ejaculation. Asian J Androl 2004;6:121–6.
[11] Symonds T, Roblin D, Hart K, et al. How does premature
ejaculation impact a man’s life? J Sex Marital Ther
2003;29:361–70.
[12] Althof S, Rosen R, Symonds T, et al. Development and
validation of new questionnaire to assess sexual satisfac-
tion, control, and distress associated with premature
ejaculation. J Sex Med 2006;3:465–75.
[13] Waldinger MD, Hengeveld MW, Zwinderman AH, et al. An
empirical operational study of DSM-IV diagnostic criteria
for PE. Int J Psychiatry Clin Prac 1998;2:287–93.
[14] Waldinger MD, Quinn P, Dilleen M, et al. A multinational
population survey of IELT. J Sex Med 2005;2:492–7.
european urology 52 (2007) 565–573 573